Glial tumors are an important clinical problem of both adults and children. The development of successful strategies for the prevention and treatment of these tumors has been elusive, in part as a result of our inability to properly classify these tumors as homogeneous biological entities. The development of markers which provide insights into pathogenesis, enhance patient management, allow for accurate prognostication, facilitate the effective grouping of patients for treatment and treatment evaluation is a central goal of clinical oncology. Cancer is the composite of pathologic behaviors that constitute what we recognize clinically as malignancy. We believe that the panoply of biologic activities that distinguish a tumor cell from its corresponding normal cell type is best explained by thinking of this composite as originating not only in specific genetic alterations that modify a particular biologic pathway, but also as the result of the cell mimicking the behavior of an embryonic cell and expressing some of those cellular characteristics that are both physiologic and appropriate for a cell type that exists during the normal ontogenesis of the lineage in which a particular tumor arises. This hypothesis is the foundation upon which our approach to the development of new tumor markers has been built. In this application we have proposed two different strategies for identifying new glial tumor markers: screening for genes that are differentially expressed in different situations thought to contrast various biologic features of pathologic significance (e.g. invasive vs. non-invasive) or screening for genomic regions that are either deleted or amplified in specific subgroups of glial tumors. In these experiments we will use state-of-the-art technologies including subtractive hybridization cloning, differential display of mRNA, and comparative genomic hybridization. We anticipate that candidate markers will be readily identified. To properly characterize and develop these candidates as effective tumor markers we have identified one possible algorithm, and we have identified selected reagents that can be efficaciously evaluated as part of the Glial Tumor Network's activities to characterize specific markers. These reagents include the pathologic specimens of entire cadres of glial tumor patients homogeneously-treated on previous therapeutic protocols at UCSF and patients participating in a large epidemiological study of brain tumors in Northern California. For all these patients computerized databases of information exist that will facilitate the critical evaluation of individual markers.